The influence of nuclearity and charge of chiral Rh diene complexes on the activity and enantioselectivity in catalytic asymmetric 1,2-additions of organoboron reagents to Ntosylimines and 1,4-additions to enones was investigated. For this purpose, cationic dimeric Rh(I) complex [(Rh(1)) 2 Cl]SbF 6 and cationic monomeric Rh(I) complex [RhOH 2 (2)]SbF 6 were synthesized from oxazolidinone-substituted 3-phenylnorbornadiene ligands 1 and 2, which differ in the substitution pattern at oxazolidinone C-5′ (CMe 2 vs CH 2 ) and compared with the corresponding neutral dimeric and monomeric Rh(I) complexes [RhCl(1)] 2 and [RhCl(2)]. Structural, electronic, and mechanistic insights were gained by X-ray crystallography, cyclic voltammetry (CV), X-ray absorption spectroscopy (XAS), and DFT calculations. CV revealed an increased stability of cationic vs neutral Rh complexes toward oxidation. Comparison of solid-state and solution XAS (extended X-ray absorption fine structure (EXAFS), X-ray absorption near edge structure (XANES)) data showed that the monomeric Rh complex [RhCl(2)] maintained its electronic state and coordination sphere in solution, whereas the dimeric Rh complex [RhCl(1)] 2 exchanges bridging chloro ligands by dioxane molecules in solution. In both 1,2-and 1,4-addition reactions, monomeric Rh complexes [RhCl(2)] and [RhOH 2 (2)]SbF 6 gave better yields as compared to dimeric complexes [RhCl(1)] 2 and [(Rh(1)) 2 Cl]SbF 6 . Regarding enantioselectivities, dimeric Rh species [RhCl(1)] 2 and [(Rh(1)) 2 Cl]SbF 6 performed better than monomeric Rh species in the 1,2-addition, while the opposite was true for the 1,4-addition. Neutral Rh complexes performed better than cationic complexes. Microemulsions improved the yields of 1,2-additions due to a most probable enrichment of Rh complexes in the amphiphilic film and provided a strong influence of the complex nuclearity and charge on the stereocontrol. A strong nonlinear-like effect (NLLE) was observed in 1,2-additions, when diastereomeric mixtures of ligands 1 and epi-1 were employed. The pronounced substrate dependency of the 1,4-addition could be rationalized by DFT calculations.

Three-dimensional (3D) organoid culture recapitulating patient-specific histopathological and molecular diversity offers great promise for precision medicine in cancer. In this study, we established label-free imaging procedures, including Raman microspectroscopy (RMS) and fluorescence lifetime imaging microscopy (FLIM), for in situ cellular analysis and metabolic monitoring of drug treatment efficacy. Primary tumor and urine specimens were utilized to generate bladder cancer organoids, which were further treated with various concentrations of pharmaceutical agents relevant for the treatment of bladder cancer (i.e., cisplatin, venetoclax). Direct cellular response upon drug treatment was monitored by RMS. Raman spectra of treated and untreated bladder cancer organoids were compared using multivariate data analysis to monitor the impact of drugs on subcellular structures such as nuclei and mitochondria based on shifts and intensity changes of specific molecular vibrations. The effects of different drugs on cell metabolism were assessed by the local autofluorophore environment of NADH and FAD, determined by multiexponential fitting of lifetime decays. Data-driven neural network and data validation analyses (k-means clustering) were performed to retrieve additional and non-biased biomarkers for the classification of drug-specific responsiveness. Together, FLIM and RMS allowed for non-invasive and molecular-sensitive monitoring of tumor-drug interactions, providing the potential to determine and optimize patient-specific treatment efficacy.

<p>Selecting only the relevant subsets from all gathered data has never been as challenging as it is in these times of big data and sensor fusion. Multiple complementary methods have emerged for observing similar phenomena and oftentimes many of these techniques are superimposed in order to make the best possible decisions. A pathologist for example uses microscopic and spectroscopic techniques for the discrimination between healthy and cancerous tissue. Especially in the field of spectroscopy, immensely many frequencies are recorded and appropriately sized datasets are rarely acquired due to time intensive measurements and the lack of patients. In order to cope with the curse of dimensionality in machine learning, it is necessary to reduce the overhead from irrelevant or redundant features.</p> <p><br></p> <p>In this article, we propose the FeaSel algorithm, that can be embedded in conventional neural networks. It recursively prunes the input nodes after the optimizer in the neural network achieves satisfying results. The weights of nodes that do not contribute critical information for the decision making will be deleted during the pruning process. We demonstrate the performance of the feature selection algorithm on different datasets and compare it to existing feature selection methods. Our algorithm combines the advantages of neural networks' non-linear learning ability and the embedding of the feature selection algorithm into the actual classifier optimization.</p>

Selecting only the relevant subsets from all gathered data has never been as challenging as it is in these times of big data and sensor fusion. Multiple complementary methods have emerged for the observation of similar phenomena; oftentimes, many of these techniques are superimposed in order to make the best possible decisions. A pathologist, for example, uses microscopic and spectroscopic techniques to discriminate between healthy and cancerous tissue. Especially in the field of spectroscopy in medicine, an immense number of frequencies are recorded and appropriately sized datasets are rarely acquired due to the time-intensive measurements and the lack of patients. In order to cope with the curse of dimensionality in machine learning, it is necessary to reduce the overhead from irrelevant or redundant features. In this article, we propose a feature selection callback algorithm (FeaSel-Net) that can be embedded in deep neural networks. It recursively prunes the input nodes after the optimizer in the neural network achieves satisfying results. We demonstrate the performance of the feature selection algorithm on different publicly available datasets and compare it to existing feature selection methods. Our algorithm combines the advantages of neural networks’ nonlinear learning ability and the embedding of the feature selection algorithm into the actual classifier optimization.

No abstract

Differentiation of malign and benign tissue based on spectral information can be done by only using a small fractional amount of the original mid-infrared spectrum. An optimally selected arrangement of a few narrow-band quantum cascade lasers provides proficient signal-to-noise ratios and can drastically reduce the data acquisition time with constant discriminability, such that real-time applications will be possible in short-term and in-vivo diagnostics in the long-term.

Infrared spectroscopy is often used to spot differences between benign and malignant tissue. Due to the proliferation of tumorous cells, the composition of tissue changes drastically. In the consequence shifts occur in its optical properties that are indicated by spectral biomarkers in the so-called fingerprint region. In this work, we propose a new concept for a sparsified multi-spectral measurement of the most important and informative biomarker signals. The results of a data-driven feature selection approach show that a reliable discrimination of the tissue is still possible, even though utilizing only a small fraction of the measured data. A selected arrangement of only a few narrow-band quantum cascade lasers could provide proficient signal-to-noise ratios and can noticeably reduce the data acquisition time. Consequentially, real-time applications will be possible in short-term and in-vivo diagnostics in the long-term. First measurements of silicone phantoms validate the imaging capability of the sensor concept.

<p>Selecting only the relevant subsets from all gathered data has never been as challenging as it is in these times of big data and sensor fusion. Multiple complementary methods have emerged for observing similar phenomena and oftentimes many of these techniques are superimposed in order to make the best possible decisions. A pathologist for example uses microscopic and spectroscopic techniques for the discrimination between healthy and cancerous tissue. Especially in the field of spectroscopy, immensely many frequencies are recorded and appropriately sized datasets are rarely acquired due to time intensive measurements and the lack of patients. In order to cope with the curse of dimensionality in machine learning, it is necessary to reduce the overhead from irrelevant or redundant features.</p> <p><br></p> <p>In this article, we propose the FeaSel algorithm, that can be embedded in conventional neural networks. It recursively prunes the input nodes after the optimizer in the neural network achieves satisfying results. The weights of nodes that do not contribute critical information for the decision making will be deleted during the pruning process. We demonstrate the performance of the feature selection algorithm on different datasets and compare it to existing feature selection methods. Our algorithm combines the advantages of neural networks' non-linear learning ability and the embedding of the feature selection algorithm into the actual classifier optimization.</p>